Mungyu Bae

Preserving privacy and efficiency in data communication and aggregation for AMI network

Due to the evolution of smart grid technologies, the number of smart meters deployed in a consumer area is rapidly increasing. As a consequence, network traffic issues in advanced metering infrastructure (AMI) have been raised. In smart gird, satisfying latency requirement and privacy-preserving are important issues. Intuitively, latency requirement can be met by reducing network delay and processing time. However, preserving privacy and security of user data has negative effects on both the network delay and the processing time due to the increased message size and cryptographic overhead. To overcome these problems, various types of message aggregation schemes with cryptographic algorithms have been proposed, but they have limitation in processing time and applicability to diverse scenarios. In this paper, we propose a data communication scheme for AMI network that can reduce the network traffic and the processing time while preserving each users privacy. With the proposed scheme, the cryptographic overhead incurred during message aggregation can be significantly reduced compared to other existing schemes. We model the network characteristic, analyze the complexity of the proposed scheme, and simulate the proposed scheme under diverse topologies and application scenarios. Through the results, we show that the proposed scheme has lower computational overhead than other schemes, and it can be applied to various smart grid applications. HighlightsWe propose a novel data communication and aggregation scheme for AMI network.The message structure for efficient and secure communication is proposed.We introduce a novel homomorphic encryption algorithm suitable to AMI.Complexity of the proposed scheme is analyzed and compared with other schemes.We evaluate performance of the proposed scheme under diverse application scenarios.

FASE: Fast authentication system for e-health

For the past few decades, it is common to observe the patients paper work at the local hospital. Even though the same personal information is used, a novel way to actually decrement the amount of these paper works does not exist. In this paper, we introduce QR code based e-health authentication system to reduce the redundant paper work and obtain patients health record easily and securely in the local hospital. With our proposed scheme, we believe that it will improve efficiency in terms of the cost and time for both patient and the hospital and protect patients personal information.

Devising Mobile Sensing and Actuation Infrastructure with Drones

Vast applications and services have been enabled as the number of mobile or sensing devices with communication capabilities has grown. However, managing the devices, integrating networks or combining services across different networks has become a new problem since each network is not directly connected via back-end core networks or servers. The issue is and has been discussed especially in wireless sensor and actuator networks (WSAN). In such systems, sensors and actuators are tightly coupled, so when an independent WSAN needs to collaborate with other networks, it is difficult to adequately combine them into an integrated infrastructure. In this paper, we propose drone-as-a-gateway (DaaG), which uses drones as mobile gateways to interconnect isolated networks or combine independent services. Our system contains features that focus on the service being provided in the order of importance, different from an adaptive simple mobile sink system or delay-tolerant system. Our simulation results have shown that the proposed system is able to activate actuators in the order of importance of the service, which uses separate sensors’ data, and it consumes almost the same time in comparison with other path-planning algorithms. Moreover, we have implemented DaaG and presented results in a field test to show that it can enable large-scale on-demand deployment of sensing and actuation infrastructure or the Internet of Things (IoT).

Wi-Fi Seeker: A link and Load Aware AP Selection Algorithm

The demand of mobile traffic increases tremendously as various Internet services are provided to mobile phones and computing devices, making mobile carriers increase the capacity of their cellular network. Since current cell towers and auxiliary devices cannot carry an excessive amount of traffic, several mobile carriers try to install low cost wireless equipment such as a Wi-Fi access point (AP) in order to offload overloaded traffic on cell towers. However, APs are installed randomly unlike cell towers because they can be installed by anyone and anywhere. The randomized allocation of APs incurs interference that causes serious degradation of wireless communication quality. Therefore, it is necessary to design mobile devices to use the environment or channel that contains the least interference for proper Wi-Fi communication. In order to make a mobile device experience the best wireless download quality, this paper proposes a novel AP selection mechanism that selects an AP that contains the least interference level compared to others with different number of communicating devices and amount of traffic load. The proposed selection mechanism does not require any additional feedback from nearby devices or APs. Performance evaluation was done by using real mobile devices with open source platform called Android.

Avss: Airborne video surveillance system

Most surveillance systems only contain CCTVs. CCTVs, however, provide only limited maneuverability against dynamic targets and are inefficient for short term surveillance. Such limitations do not raise much concern in some cases, but for the scenario in which traditional surveillance systems do not suffice, adopting a fleet of UAVs can help overcoming the limitations. In this paper, we present a surveillance system implemented with a fleet of unmanned aerial vehicles (UAVs). A surveillance system implemented with a fleet of UAVs is easy to deploy and maintain. A UAV fleet requires little time to deploy and set up, and removing the surveillance is also virtually instant. The system we propose deploys UAVs to the target area for installation and perform surveillance operations. The camera mounted UAVs act as surveillance probes, the server provides overall control of the surveillance system, and the fleet platform provides fleet-wise control of the UAVs. In the proposed system, the UAVs establish a network and enable multi-hop communication, which allows the system to widen its coverage area. The operator of the system can control the fleet of UAVs via the fleet platform and receive surveillance information gathered by the UAVs. The proposed system is described in detail along with the algorithm for effective placement of the UAVs. The prototype of the system is presented, and the experiment carried out shows that the system can successfully perform surveillance over an area set by the system.

Alarm system for elder patients medication with IoT-enabled pill bottle

Nowadays, Internet of Things (IoT) is deeply involved in human lives. Particularly, when IoT is applied to medical service, it can greatly benefit people. This is especially true for people who cannot take care of themselves and need constant care, such as the elderly. As the number of elderly people living alone has increased, home caring services have become necessary. As a service to be used among those, we propose self-alarm system for medication. For many patients, it is easy to forget to take their medicine, which is an important element for treatment. We track state changes of medicine bottles by using weight sensors. According to the tracking record, our system will decide whether or not to alarm the patient. As a result, patients are able to take their medicine on time.

MACoD: Mobile Device-Centric Approach for Identifying Access Point Congestion

These days, since many mobile devices support Wi-Fi, the number of Wi-Fi hotspots are increasing. At the same time, the number of places where users can use wireless LAN are increasing. Unlike the cellular network, the traffic of wireless LAN may not be distributed evenly because there is no supervisor to control the network traffic. In addition, due to the low throughput and the high packet error rate in wireless communication, the network traffic may converge at some Access Points (APs) which take a role of connecting wired and wireless communication interfaces. Therefore, congestion can easily happen at APs in wireless LAN. However, received signal strength indicator (RSSI) frequently used does not mean the network state of AP, and existing sender-based congestion detection and control protocols cannot recognize accurately where the congestion occurs. Therefore, current mobile devices continuously use the connected AP even if it is congested after the association. In this paper, we propose MACoD, the low power AP congestion detection system for mobile devices, in order to resolve aforementioned problems. Furthermore, by conducting diverse simulations, actual experiments, and power consumption analysis, we validate that MACoD can detect congestion effectively and that it is suitable for mobile devices.

A simulation study on Map/Reduce framework in wireless data center environment

Big data has different characteristics from the traditional data. It has large volume and various type and requires real-time processing. To process the big data, cluster based parallel and distributed computing model, Map/Reduce has been introduced. The Map/Reduce is basically designed for the cluster where the processing units are connected to each other with wired network, called data center. However, in the future, the wireless data center would be introduced since it has low installation and management complexity and cost. To introduce the wireless data center, we need to answer the fundamental question, “How does Map/Reduce framework would operate in the wireless data center?” To answer the question, in this paper, we perform simulation study of the Map/Reduce application in the wireless environment. We introduce the basic simulation model and the results of the simulation study to verify the potential of the wireless data center.

RSSI localization with DB-Assisted Least Error algorithm

RSSI (Received Signal Strength Indication) localization techniques using Wi-Fi presents substantial advantages compared to others. They are light weight both in terms of computation and energy consumption. RSSI localization techniques are mostly used indoors, where many APs (Access Points) are present and no GPS is available. Recently, APs are getting deployed outdoors as well, and urban canyon phenomenon degrades the capability of GPS localization even in outdoor environments, which makes RSSI localization techniques attractive as an outdoor localization solution as well. The downside of RSSI localization is that it is polarized, which means it has either high performance and economic cost or low cost and poor accuracy. Both cases are inadequate for a general deployment; high-cost algorithms can only be deployed in heavily populated area for cost feasibility and the accuracy of low-cost algorithms is nowhere near credible. In this paper, we propose a range-based RSSI localization algorithm that has reasonable accuracy yet has very low cost. The proposed algorithm consists of DB-assistance, ration base algorithm, and an elementary machine learning algorithm. This helps achieving the qualities that can provide a feasible RSSI localization solution that can be employed in a much wider area.

Link-aware AP selection for improving Wi-Fi quality

The demand of mobile traffic increases tremendously as various Internet services are provided to mobile phones and computing devices, making mobile carriers to increase capacity of cellular network. Since current cell towers and auxiliary devices cannot carry excessive amount of traffic, several mobile carriers try to install low cost wireless equipments such as Wi-Fi access points (AP) so that some part of mobile traffic can be offloaded to these wireless nodes. Unlike cell towers, APs are installed randomly since there is no specific allocation policy for these license-exempt band equipment that are open to everyone. Since the randomized allocation of APs incur interference that causes serious degradation of wireless communication quality, it is necessary to design mobile devices to communicate in the environment or channel that contains the least interference. In order to make a mobile device experience the best wireless download quality, this paper proposes a novel AP selection mechanism that selects an AP that contains the least interference level compared to others with different number of communicating devices and amount of traffic load. Proposed selection mechanism does not require any additional feedback from nearby devices or APs. Performance evaluation is done by using real mobile devices with open source platform called Android.